L14 Tables 1

Data Visualization (STAT 302)

Overview

The goal of this lab is to explore static tables using the grammar of tables from the gt and gtExtras package.

The gt() function creates a gt table object when provided with table data. Using this function is the first step in a typical gt workflow. Once we have the gt table object, we can perform styling transformations by adding on column, formatting, tab styling, and summary layers.

A few helpful links to provide more information about the gt package: gt rstudio and intro to gt

Posit (formally RStudio) hosts an annual table competition! Check out some past winning tables: 2021 2022

Dataset

We will be using the pizzaplace dataset included in the gt package, the house_prices, stream_data, and team_logo dataset which should be located in your /data subdirectory. Each dataset will require a bit of data wrangling provided in their respective exercise code chunk.

# load package(s)
library(tidyverse)
library(gt)
library(gtExtras)

# load data
# exists w/ gt library - data() puts it into environment
data("pizzaplace")
# load paces and stores rda file in environment
load("data/house_prices.rda")
# reads data, the arrow stores it in environment
stream_data <- read_csv("data/stream_data.csv", show_col_types = FALSE) |> 
  janitor::clean_names()
# similar to csv
team_logo <- read_rds("data/team_logo.rds") |> 
  janitor::clean_names()

Exercise 1

Use the pizza_summary dataset created below to recreate the following graphic as precisely as possible. We will work through this exercise together to provide a high level overview of a few of the ways you can layer on table elements to make a simple but visually appealing table.

Hints:

• md(): Interpret input text as Markdown-formatted text

Code

# data wrangling
pizza_summary <- pizzaplace |> 
  filter(type %in% c("classic", "chicken")) |> 
  group_by(name, type) |> 
  summarize(
    number_sold = n(),
    profit = sum(price)
    ) |> 
  ungroup()

Solution

Code

# build table
pizza_summary |> 
  gt(
    rowname_col = "name",
    groupname_col = "type"
    
  ) |> 
  tab_header(
    title = md("**Pizza Place Annual Pizza Sales**"),
    subtitle = "Fiscal Year 2015"
  ) |> 
  tab_source_note(
    source_note = md("Source: *gt* package dataset")
  ) |> 
  cols_label(
    # variable = "string"
    number_sold = md("Number<br>Sold"),
    profit = md("Annual<br>Revenue")
  ) |> 
  # reorders rows, not rename
  row_group_order(groups = c("classic", "chicken")) |> 
  fmt_currency(
    columns = profit,
    currency = "USD",
    decimals = 2
  ) |> 
  fmt_number(
    columns = number_sold,
    decimals = 0
  ) |> 
  tab_style_by_grp(
    column = profit,
    fn = max,
    cell_fill(color = "red", alpha = 0.25)
  ) |> 
  tab_style_by_grp(
    column = number_sold,
    fn = max,
    cell_fill(color = "blue", alpha = 0.25)
  ) |> 
  tab_footnote(
    footnote = "Our most popular pizza",
    locations = cells_stub("classic_dlx")
  ) |> 
  tab_options(
    row_group.background.color = "gray40",
    stub.border.style = "dashed"
  ) |> 
  opt_footnote_marks(
    marks = "standard"
  ) |> 
  opt_table_outline()

Pizza Place Annual Pizza Sales
Fiscal Year 2015
	Number Sold	Annual Revenue
classic
big_meat	1,914	$22,968.00
classic_dlx*	2,453	$38,180.50
hawaiian	2,422	$32,273.25
ital_cpcllo	1,438	$25,094.00
napolitana	1,464	$24,087.00
pep_msh_pep	1,359	$18,834.50
pepperoni	2,418	$30,161.75
the_greek	1,420	$28,454.10
chicken
bbq_ckn	2,432	$42,768.00
cali_ckn	2,370	$41,409.50
ckn_alfredo	987	$16,900.25
ckn_pesto	973	$16,701.75
southw_ckn	1,917	$34,705.75
thai_ckn	2,371	$43,434.25
Source: gt package dataset
* Our most popular pizza

Exercise 2

Using the pizza_size dataset derived from the pizzaplace dataset, recreate the following graphic as precisely as possible.

Code

# data wrangling
pizza_size <- pizzaplace |> 
  filter(
    size %in% c("S", "L", "M"), 
    type %in% c("chicken", "classic", "veggie")
    ) |> 
  group_by(type, size) |> 
  summarize(
    number_sold = n(),
    profit = sum(price)
    ) |> 
  ungroup() |> 
  rename(category = type)

Table 1

Hints:

• tab_style() is useful for changing cell colors
• Hex codes: "#F8766D", "#00BA38", "#619CFF"
• size L: bold
• size M: italic
• size S: "white"

Solution

Code

base_pizza <- pizza_size |> 
  gt(
    # what each row w/in a group will be
    rowname_col = "size",
    # the groups that rowname_col will be for each
    groupname_col = "category"
  ) |> 
  tab_header(
    title = md("**Pizza Place Annual Pizza Sales**"),
    subtitle = "Fiscal Year 2015"
  )

Code

base_pizza |> 
  # have to do a different tab_style for each row
  tab_style(
    style = list(
      cell_fill(color = "#F8766D"),
      cell_text(color = "white")),
    locations = cells_body(rows = size =="S")
  ) |> 
  tab_style(
    style = list(
      cell_fill(color = "#00BA38"),
      cell_text(color = "black", style = "italic")),
    locations = cells_body(rows = size =="M")
  ) |> 
  tab_style(
    style = list(
      cell_fill(color = "#619CFF"),
      cell_text(color = "black", weight = "bold")),
    locations = cells_body(rows = size =="L")
  )

Pizza Place Annual Pizza Sales
Fiscal Year 2015
	number_sold	profit
chicken
L	4932	102339.00
M	3894	65224.50
S	2224	28356.00
classic
L	4057	74518.50
M	4112	60581.75
S	6139	69870.25
veggie
L	5403	104202.70
M	3583	57101.00
S	2663	32386.75

Table 2

Hints:

• gt_color_rows is useful for changing entire column colors
• manual colors c("white", "skyblue")
• Pastel1 palette from RColorBrewer package

Solution

Code

base_pizza |> 
  gt_color_rows(
    columns = number_sold,
    palette = RColorBrewer::brewer.pal(n = 9, name = "Pastel1")
  ) |> 
  gt_color_rows(
    columns = profit,
    palette = c("white", "skyblue")
  )

Pizza Place Annual Pizza Sales
Fiscal Year 2015
	number_sold	profit
chicken
L	4932	102339.00
M	3894	65224.50
S	2224	28356.00
classic
L	4057	74518.50
M	4112	60581.75
S	6139	69870.25
veggie
L	5403	104202.70
M	3583	57101.00
S	2663	32386.75

Exercise 3

Using the house_illinois dataset derived from the house_price dataset, recreate the following graphic as precisely as possible.

Similar to the ggplot scales_*() function, the gt package has useful transformations that can be performed using the following syntax tab_*(), fmt_*(), cols_*(). Check the package documentation for a full list of options.

Code

# data wrangling
house_illinois <- house_prices |> 
  filter(state %in% c("Illinois")) |> 
  mutate(year = year(date)) |> 
  filter(year %in% c(2005:2010) ) |> 
  group_by(year) |> 
  summarize(
    hpi = mean(house_price_index),
    change = mean(house_price_perc)
    )

Hints:

• color is "red" for all percents less than 0
• percent is accounting
• currency "USD"
• "Housing Price Index" label is a tab_*() layer

Solution

Code

house_illinois |> 
  gt(
    # this removes the label of the column
    rowname_col = "year"
    ) |> 
  tab_header(
    title = "Average Annual Housing Price Index (HPI)",
    subtitle = md("*Illinois Prices for 2006-2010*")
  ) |> 
  tab_style(
    style = cell_text(color = "red"),
    location = cells_body(
      columns = change,
      # this is the check for if it's red or not
      rows = change < 0
    )
  ) |> 
  fmt_currency(
    columns = hpi,
    currency = "USD",
    decimals = 2
  ) |> 
  fmt_percent(
    columns = change,
    decimals = 2,
    # gives the same number of significant digits
    scale_values = TRUE,
    # commas and () for negative
    accounting = TRUE
  ) |> 
  tab_spanner(
    # the name of the new column group
    label = "Housing Price Index",
    # the columns the group is made for
    columns = c(hpi, change)
  ) |> 
  cols_label(
    hpi = "Value",
    change = "% Change"
  ) |> 
  tab_footnote(
    footnote = "Housing market crash",
    locations = cells_stub(rows = "2008")
  )

Average Annual Housing Price Index (HPI)
Illinois Prices for 2006-2010
	Housing Price Index
	Value	% Change
2005	$138.39	7.81%
2006	$145.69	5.31%
2007	$145.54	(0.09%)
1 2008	$135.06	(7.22%)
2009	$122.30	(9.40%)
2010	$116.27	(4.93%)
1 Housing market crash

Exercise 4

Adding icons and graphics to a table is made easy with gtExtras. Using the stream_data dataset, recreate the following graphic as precisely as possible.

Hints:

• gt_plt_bullet()
• gt_fa_column()
• nytimes theme()
• fmt_symbol_first()

Solution

Code

# pipe the following code into your gt functions
stream_data |> 
  mutate(
    type = map(type, fontawesome::fa)
  ) |>
  gt() |> 
  gt_theme_nytimes() |> 
  gt_plt_bullet(
    # column is the bar
    column = nominee,
    # target is the actual bullet value that it would put on top of nominee
    target = winner,
    # pallette = c(bar color, bullet color). default is grey and red
    palette = c("grey", "blue")
  ) |> 
  fmt_symbol_first(
    column = ratio,
    suffix = "%",
    decimals = 1,
    scale_by = 100
  ) |> 
  tab_header(
    title = md("**Netflix leads Emmy Nominations & Wins:<br>2020-2021**"),
    subtitle = md("but **HBO** leads in their Wins to Nominations ratio")
  ) |> 
  gt_highlight_rows(
    # this tells which variable it looks through, then it looks through distributor for HBO and highlights that
    rows = distributor == "HBO",
    fill = "grey",
    alpha = 0.4
  ) |> 
  gt_add_divider(
    columns = type,
    sides = "right",
    color = "lightgrey",
    weight = px(1)
  ) |> 
  cols_width(
    distributor ~ px(100),
    type ~ px(40),
    ratio ~ px(60),
    nominee ~ px(230)
  )

Netflix leads Emmy Nominations & Wins: 2020-2021
but HBO leads in their Wins to Nominations ratio
distributor	type	ratio	nominee
Netflix		27.7%
HBO		28.8%
NBC		20.5%
ABC		16.0%
Hulu		2.0%
FX Networks		11.1%
Prime Video		8.9%
Apple TV+		23.8%
CBS		2.5%

Exercise 5

Using the team_logo dataset, recreate the following graphic as precisely as possible.

The tab_options() layer for gt is similar to the theme() option for ggplot. Here you can add customization to the background colors, line type, size, etc.

Hints:

• gt_img_rows with height = 30
• tab options:
  • data row padding is px(1)
  • table background color is "gray90"
  • table width is 100%
  • table font color is "red"
  • column labels have background color "black", font size "large", and font weight "bold",
  • table body has hlines color "red"
  • table body has border bottom color "red"

Solution

Code

# pipe the following code into your gt functions
team_logo |>
  dplyr::select(logo = team_logo_espn, team_name, team_nick, team_conf) |>
  slice_sample(n = 2) |> 
  gt() |> 
  gt_img_rows(
    columns = logo,
    img_source = "web",
    height = 30
  ) |> 
  tab_options(
    data_row.padding = px(1),
    table.width = "100%",
    table.background.color = "gray90",
    table.font.color = "red",
    column_labels.background.color = "black",
    column_labels.font.size = "large",
    column_labels.font.weight = "bold",
    table_body.hlines.color = "red",
    table_body.border.bottom.color =  "red"
  )

logo	team_name	team_nick	team_conf
	New England Patriots	Patriots	AFC
	New York Giants	Giants	NFC